Background Anemia and atrial fibrillation (AF) are common among the elderly. Anemia is an independent predictor of mortality and morbidity for numerous cardiovascular and noncardiovascular diseases, but the association of anemia with mortality and hospitalizations in patients with AF requires clarification. Methods Subjects were 13,0.67 Medicare beneficiaries hospitalized with AF and included in the National Registry of Atrial Fibrillation II data set. Index hospitalization hematocrit (Hct) was obtained by structured chart abstraction. Cox proportional hazards models quantified the association of Hct with mortality and re-hospitalizations during a median follow-up period of 12 months. Results The mean age was 79.8 years, 58% were women, and the mean Hct was 39.2%. Hematocrit was significantly (P < .0001) associated with risk of death and of rehospitalization even after adjustment for demographic information, comorbid conditions, and use of cardiovascular medications. As compared to a Hct of 40% to 44.9%, the adjusted hazard ratios for mortality were 1.66 for Hct <25%, 1.50 for 25% to 29.9%, 1.28 for 30% to 34.9%, 1.07 for 35% to 39.9%, 1.03 for 45% to 49.9%, and 1.10 for ≥50%. The association between anemia and mortality was significant in men and women but stronger in men (P = .006 for interaction). Compared to the category 40% to 44.9%, the risk of rehospitalization was increased to 28% (adjusted hazard ratio 1.28, 95% CI 1.15-1.43) in the Hct category 25% to 29.9%. Conclusion Anemia is an independent predictor of mortality and of hospitalizations in elderly patients with AF. Studies are needed to assess the effect of treatment of anemia on clinical outcomes.